Brick by Brick: Metformin for Gestational Diabetes Mellitus?

Jean-Luc Ardilouze; Masoud Mahdavian; Jean-Patrice Baillargeon


Expert Rev Endocrinol Metab. 2010;5(3):353-357. 

In This Article


The Balani et al. study shows that GDM outcomes for women treated with metformin are more favorable than those recorded for women treated with insulin.

In mothers, blunting of weight gain by metformin was expected and this effect was also reported in the MiG trial.[11] The nonsignificant difference in HDP supports metformin use in GDM. Indeed the MiG study reported an increase of HDP in the metformin group, a finding that was quite surprising since a large body of evidence suggests that HDP and GDM share a common insulin resistance background.[16,17] The findings of the Balani et al. paper thus support this rationale.

Other results that lend support to the use of metformin during pregnancy were also found in the Balani et al. study. Only 8% of subjects required insulin to reach glucose targets. This is far from the 43% observed in the MiG trial. The possible effects on the results of ethnicity and/or weight, if any, are unknown. Roughly half of the participants in the Balani et al. study were Caucasian and the other half Asian or African, whereas 55% of participants in the MiG trial were Polynesian, Indian and Chinese.[11] Mean BMI in the Balani et al. study was of 30.4 versus 32.0 kg/m2 in the MiG trial.[11] Confirming these data is a report of a randomized study comparing metformin and glyburide in 150 subjects.[18] Results showed that mean fasting and 2-h postprandial blood glucose levels were similar in both groups, although 26 patients in the metformin group and 12 patients in the glyburide group did not achieve adequate glycemic control and required insulin therapy (34.7 vs 16.2%; p = 0.01). These data thus concur with the glyburide failure rate (4%) observed by Langer et al..[12] OHA failures provide basis to the claim that, while metformin and glyburide are clinically effective alternatives to insulin therapy, the latter should be favored because it provides adequate control of blood glucose in a greater proportion of women with GDM than do the former. However, this claim conveniently disregards women's attitudes toward treatment options. Indeed, the MiG trial reported that more women in the metformin group than in the insulin group (76.6 vs 27.2%; p < 0.001) would choose to receive their assigned treatment again.[11]

The prevalence of prematurity was even lower with metformin in the Balani et al. study than in the MiG study. Lower mean birth weight centile for gestational age in the metformin group and the decrease in incidence of macrosomia are also data reflecting the potential benefits of metformin as an insulin sensitizer that crosses the placenta.

Nine participants were excluded for 'noncompliance' but, according to the authors, this noncompliance was in fact subjects' refusal to take a drug not yet licensed for use in pregnancy. This is also significant since OHA are not yet universally recommended for use during pregnancy. Professional associations and national institutes remain prudent.[19,101] Indeed, the Australasian Diabetes Association as well as the American Diabetes Association are still hesitant toward the use of OHA in pregnancy. Those that do allow their use suggest careful management, prior consent from patients, strict assessment of neonates and long-term monitoring of offspring. While certain associations have somewhat opened-up to the prescription of OHA since 2008, the Canadian Diabetes Association (CDA) recommendations are to the effect that "glyburide or metformin may be considered in women with GDM who are non-adherent to or who refuse insulin". The CDA specifies that the use of metformin or glyburide during pregnancy is currently not an approved indication in Canada, and that such use would be considered off-label.[20] In 2008, the England and Wales National Institute for Health and Clinical Excellence also included metformin and glyburide as GDM and Type 2 diabetes in pregnancy treatment options, with the proviso that it is not licensed for these indications.[101] In 2009, the International Diabetes Federation (IDF) stated that no teratogenic problems related to metformin had been found, although this question has not been systematically addressed, and the MiG study had shown metformin to be a viable alternative to insulin.[21]

The teratogenic effect of OHA leading to congenital malformation has been addressed in the literature reporting on treatment of GDM. However, the evidence in support of such an effect is weak. Congenital malformations were reported, principally, in a series of historic cases of truly hyperglycemic pregnant women treated with first-generation sulfonylureas and biguanides. Hyperglycemia per se is now considered teratogenic. Recent meta-analyses[22] and a Canadian statement[23] reported on the absence of any teratogenic effect of OHA. The Balani et al. results are in line with this point of view as no deleterious effect of metformin was recorded. On the contrary, the mean birth weight centile of neonates in the metformin group was reduced by 20% compared with the insulin group and cases of jaundice and hypoglycemia were fewer, as were admissions to the intensive care unit.

Among the possible limitations of the Balani et al. report, one could argue that the small size of the study forgoes assessment of less frequent secondary outcomes. In addition, the exact duration of treatment is unclear. It seems that all women were not recruited following the recommended 24–28-week gestation OGTT test as claimed by the authors; indeed the range for treatment with metformin and insulin extended to 25 and 24 weeks respectively, implying that certain participants underwent therapies as of the 15–16th weeks. If this was the case, then this is the first study to our knowledge reporting on prolonged therapy with metformin in women with GDM, with the exception of women with polycystic ovary syndrome, in whom a prospective cohort study found that metformin prevented GDM when used throughout pregnancy.[24] In the Balani et al. paper, neither insulin titration nor HbA1c at delivery is discussed. We assume that titration was performed according to routine medical practice, using SMBG and predefined fasting and postprandial glucose targets. This data would be useful to interpret the frequency of hypoglycemia in mothers and neonates. The Balani et al. results apply only to women who tolerate metformin and achieve control of their glucose levels with metformin alone, since the study was not intent-to-treat. Furthermore, because the trial was not randomized or blinded, and since the insulin group was retrospective, it is possible that participants in the metformin group were healthier – as suggested by a lower frequency of history of previous GDM, a trend for lower glucose levels during the diagnostic OGTT and an increased gestational age at entry – and received better care. Although the aforementioned limitations may have improved the benefits of metformin, they would not have masked any significant adverse effects.

In recent years, cornerstone randomized studies showed that the treatment of even the mildest forms of GDM improves the outcomes of pregnancy.[8,9] Pathophysiology-oriented treatment of GDM has also been addressed. Langer et al. proved that glyburide overcomes insulin secretion failure, with similar glucose mean concentrations during treatment and no significant differences between glyburide and insulin groups in large for gestational age, macrosomia, hypoglycemia, admission to intensive care unit or fetal anomalies.[12] Insulin resistance was overcome by metformin in the MiG trial and severe hypoglycemia was less frequent in neonates receiving metformin, with identical incidence of fetal anomalies. Thus, metformin appears to be a safe alternative to insulin therapy. Yet metformin does cross the placenta. Results of the MiG Offspring Follow-up (TOFU) trial, expected in 2015, should provide much-needed data on metformin's long-term safety in offspring. In the meantime, we know that offspring of polycystic ovary syndrome women treated with metformin throughout pregnancy display normal growth and motor-social development in the first 18 months of life.[25]

As patients are more likely to prefer and adhere to oral treatment, such medication is highly desirable for the effective control of blood glucose during pregnancy. Should larger studies confirm results discussed above, metformin, as well as glyburide, could become choice agents for treatment of GDM.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.